LangGraph State & Persistence
Manages persistent workflow states, fault-tolerant execution recovery, and multi-threaded conversation memory for LangGraph agents.
This skill provides specialized patterns and best practices for implementing robust checkpointing in LangGraph. It enables AI agents to resume interrupted executions, manage thread-scoped short-term memory, and implement long-term cross-thread persistence using SQLite or PostgreSQL. By handling state history, graph migrations, and automatic recovery, it ensures that production workflows remain resilient and debuggable through comprehensive state inspection and rollback capabilities.
主な機能
01Graph topology migration strategies for evolving production workflows
02Automated workflow recovery and resumption via deterministic thread IDs
0369 GitHub stars
04Multi-backend support including In-memory, SQLite, and PostgreSQL savers
05State history inspection and manual rollback for debugging complex agents
06Architectural patterns for separating thread-scoped and cross-thread memory
ユースケース
01Building fault-tolerant AI agents that survive process restarts or crashes
02Creating multi-session chat applications with persistent user preferences
03Implementing human-in-the-loop workflows that pause for manual review